Aug 2026· Digital Translation· 0 citations· 22 references
TL;DR
The concept of temperature is introduced as a means for controlling stochasticity in LLM-based translation as well as a practical and replicable experiment conducted in Google Colab to establish a replicable methodological pathway.
Abstract
This article proposes a Translation Studies–oriented approach to understanding and managing variability in machine
translation outputs generated by large language models (LLMs). Drawing on the metaphor of the “stochastic parrot,” the study
introduces the concept of
temperature
as a means for controlling stochasticity in LLM-based translation. Through
a practical and replicable experiment conducted in Google Colab, technical texts are translated from English to Spanish under
varying temperature conditions. While the dataset is intentionally limited, the study’s primary contribution lies in establishing
a replicable methodological pathway rather than in producing generalizable quantitative results. By combining computational
experimentation with reflection on concepts relevant to translation theory, the study can inform both future research and
practical approaches.
Large Language Models (LLMs) are increasingly used for translation, yet their value depends on preserving meaning rather than producing fluent output. This study evaluates seven LLMs on Japanese–Croatian translation, a low-resource, typologically distant language pair. Using rubric-based human evaluation of adequacy, fluency, terminology, and register, we compare model performance. Results show a stable ranking: qwen3 performs best, followed by phi4 and gemma3, while qwen2 performs worst. Performance differences reflect structural reconstruction, particularly argument recovery, aspectual mapping, lexical precision, and register. Qualitative analysis also reveals limited differentiation within the South Slavic continuum and pragmatic inconsistencies. Although productivity effects were not measured, improved translation adequacy may reduce post-editing and verification effort.
A novel fragment-based reasoning framework is introduced in which the model first extracts parallel source-target fragments from retrieved similar exemplars, and uses these fragments as intermediate reasoning traces to produce the final translation.
Maxime Bouthors, J. Crego, François Yvon· 0 citations
Current research has demonstrated the potential of large language models, such as GPT, as powerful translation
tools. However, gaps remain in understanding how human and machine translation differ across linguistic and structural levels.
This study adopts a network-based approach, using syntactic dependency networks to investigate structural differences in
translations produced by humans and machines (Google Translate and ChatGPT). The findings revealed that human translation networks
exhibit higher clustering coefficients and shorter average path lengths compared to ChatGPT translations, along with lower density
and degree centrality than Google Translate. Human translations also contain fewer function words among central nodes than machine
translations. These findings suggest that while machine translation prioritizes producing grammatically well-formed sentences,
human translation tends to be more concise and efficient in transmitting information, optimizing the balance between syntactic
simplicity and communicative clarity. Machine translation mirrors human words but does not fully replicate the syntactic and
structural patterns of human translation. Our study offers new insights for future studies on translation in the context of
AI.
This pilot study examines how the English adverb actually is rendered in Slovak translation, drawing on a corpus of literary texts and their translations. As a multifunctional item, actually performs a range of syntactic and pragmatic functions in context, most notably those of emphasizer and disjunct. The analysis draws on three Slovak translations of works by the same English-speaking author. Although each translator worked on a different set of books, the corpus still makes it possible to compare translation strategies within a relatively consistent authorial style and to observe how individual translators deal with the same multifunctional source item. The study combines qualitative functional analysis with quantitative observations in order to identify recurrent Slovak renderings of actually, including lexical substitution, structural reformulation, attenuation, and omission. The findings suggest that translation difficulty arises not from the lack of Slovak expressive resources, but from differences between English and Slovak in the grammatical encoding of discourse-related meaning. In particular, functions expressed in English through sentence adverbials do not always correspond to a single formal category in Slovak, which increases the likelihood of functional shift or translation loss. The study contributes to translation studies and contrastive linguistics by showing how the pragmatic contribution of a frequent multifunctional item may be preserved, reinterpreted, weakened, or omitted in translation, and how these outcomes vary across translators.
Zuzana Knižková· Studies About Languages· 0 citations
It is found that LLMs can be moved toward reformulation and away from the sentence-by-sentence paradigm, though more work is needed to improve the effectiveness of those reformulations.
Light is shed on the process-oriented nature of translation strategies and empirical evidence for integrating technological tools into translation pedagogy is offered.
Xuefeng Wu, Chenchen Liu· English Language Teaching· 0 citations